The human gut harbors 10-100 trillion microorganisms that enable the harvest of nutrients/energy from otherwise undigestible components of our diet (e.g. complex plant polysaccharides). Syntrophic (cooperative) metabolism, where one microbe produces compounds that the other requires for growth or removes compounds that inhibit the progress of metabolic reactions, has a high impact on the efficiency at which our microbes extract calories from our food. Knowledge of how different human gut bacteria fit within the cascade of metabolic interactions of the anaerobic degradation process will help to relate community composition information to the function and efficiency of the gut bioreactor. The goal of this work is to use both metagenomic (16S ribosomal RNA or shotgun) sequences from human stool samples and genome sequences from cultured gut isolates to predict microbial interactions that can be further explored/verified with laboratory experiments. This work focuses on interactions between bacteria and Methanobrevibacter smithii, the most prominent archaeal methanogen in the human gut. Methanogenic archaea were chosen because 1) they can increase the efficiency of bacterial fermentation by preventing the accumulation of metabolic products such as hydrogen 2) they are thought to be a "keystone" species, (i.e. have a higher influence on community composition and function than their prevalence would suggest) and 3) they closely associate with specific syntrophic bacteria in other environments such as sludge digestors, but whether there is an analog in the gut is not known. The goal of Specific Aim 1 is to use metagenomic sequence data from human stool samples to identify species (phylotypes) and genes whose prevalence are correlated with the presence/absence of M. smithii. The preliminary analysis of 191 samples that were collected through an ongoing longitudinal study of the effects of obesity on the gut microbiota, has identified 27 bacterial phylotypes, representing at least 3 deep bacterial lineages that appear to have conserved the traits that lead to co-occurrence with M. smithii. Specific Aims 2 and 3 pursue a combination of laboratory and computational techniques to determine whether the co- occurrence between these bacteria and M. smithii is driven by syntrophy or by shared environmental preferences. Some of the co-occurring phylotypes are from uncultured lineages whose biological properties are completely unknown. Specific Aim 2 will yield information on these uncultured lineages and determine whether co-occurrence was driven by syntrophy by 1) microscopic determination of whether they form structured complexes with M. smithii using Fluorescence In Situ Hybridization (FISH) and 2) metagenomic sequencing of cell populations that were concentrated using flow cytometry. Specific Aim 3 further explores the underlying cause of co-occurrence patterns by developing and applying metabolic reconstruction-based techniques to predict interactions between microbes, including syntrophy and metabolic niche convergence. Finally, I will use this combined information to design confirmatory experiments in gnotobiotic mice. This work will facilitate the use of the growing collection of human-gut derived sequences to understand whether and how particular microbes interact, and will provide insights as to how to promote or discourage the activity M. smithii in the gut. This proposal is a natural extension of my Ph.D. and post-doctoral studies of the human microbiome. The proposed research will further develop the skills needed to achieve my goal of developing an independent research group with both computational and laboratory components. The bioinformatics work integrates my experience with analysis of 16S rRNA and genomic sequence data from the human gut, and extends my expertise into new areas, such as metabolic network modeling. The laboratory component draws upon my experience in performing culture-independent analysis of microbes in soil, and extends my training in FISH and flow-cytometry, for the generation of genomic information from uncultured microbial lineages. Extended training in working with human subjects will also help me to continue to perform human microbiome research. My current position as a post-doc with Dr. Rob Knight at the University of Colorado at Boulder, and the co-mentoring that I receive from Dr. Jeff Gordon from the Center for Genome Sciences at Washington University provide an excellent environment in which to reach these goals. The Knight lab is on the forefront of generating the computational tools required to utilize advances in high-throughput sequencing for the analysis of microbial communities, and is an environment where I can interact with a diverse collection of students, post-docs, and collaborators including individuals with backgrounds in biology (with computational and/or laboratory expertise), computer science (including high performance computing and database design), and applied math. The Gordon lab performs ground-breaking research on the association of the gut microbial community with diseases of nutrition (obesity and malnutrition) and the application of gnotobiotic mouse models to understanding microbial interactions in the gut. They produce massive amounts of sequence information from human gut samples that is central to the work proposed in this grant.